An Integrated Optimisation Method for Aluminium Hot Rolling

Ioannis Avgerinos, Apostolos Besis, Ioannis Mourtos, Athanasios Psarros, Stavros Vatikiotis, Georgios Zois

Abstract: This paper addresses the scheduling of aluminium slabs in a hot rolling mill, which is a critical stage in aluminium rolling. Hot rolling defines a challenging variant of flowshop scheduling because of multiple side constraints imposed by quality specifications plus the need for synchronization with the preceding stage where slabs are preheated in furnaces. An additional complication arises from the objective of minimizing succession penalties in the rolling mill, in order to maximise roll quality. Current practices group slabs into batches and then schedule batches to reduce succession penalties and avoid idle times during the transition from preheating to hot rolling. Motivated by the operational requirements of a leading EU manufacturer of aluminium rolled products, we propose a novel approach that replaces the conventional batching approach by treating each aluminium slab individually. The aim is to expand the range of potential successions in the hot rolling mill and thus further reduce the sum of succession penalties. To formalise a set of elaborate (hard and soft) constraints, we introduce a Mixed Integer Linear Programming model for both the hot rolling and assignment of slabs to furnaces, and a Constraint Programming model for the precise schedule computation of each slab. These models are optimised sequentially to obtain a schedule (a ‘production cycle’) given a set of available slabs. Experiments on real inventory data demonstrate that the proposed approach significantly reduces the total succession penalties, while avoiding idle times between preheating and rolling. Therefore, our study could enhance production planning and scheduling for aluminium rolling.

External IDs:dblp:conf/cpaior/AvgerinosBMPVZ25